Nonlinear modeling of the polymer Membrane Fuel Cells using Deep Belief Networks and Modified Water Strider Algorithm

Abstract

The present study presents a new method for optimal identification of the output voltage of a Proton Exchange Membrane Fuel Cell based on an optimized Deep Belief Network. According to the designed method, the Deep Belief Network structure is optimized based on a modified version of the Water Strider Algorithm. The main idea is to minimize the error value between the observed and the obtained output voltages of the Proton Exchange Membrane Fuel Cell. Simulations specified that the suggested Deep Belief Network-Modified Water Strider Algorithm method provides accurate tracking for the voltage signal prediction of the Proton Exchange Membrane Fuel Cell. Final results are also compared with Deep Belief Network-Water Strider Algorithm, and DBN/TLBO-DE from the literature to show the proposed method dominance.

Keywords

• Proton exchange membrane fuel cell
• Output voltage
• System identification
• Deep Belief Networks
• Modified Water Strider Algorithm